Re: [Getfem-users] help with MUMPS_distributed_matrix_solve

[Yves Renard]

Le mercredi 19 septembre 2007 11:56, vous avez écrit :
> hello
> i use gmm for my finite element code. I solve the linear system in my
> newton-raphson procedure with the different solver of gmm, especially
> the superlu and mumps interface. It works great, but i would like to
> use the distributed version of mumps in my code. I don't see any help
> for this interface, and i don't know how i can manage process and MPI
> with this interface.
> i call the MUMPS_distributed_matrix_solve(A,x,b) procedure directly
> in the newton subroutine.
>
> Where can i find example or documentation?

This is not documented. It is used in the file
getfem_model_solvers.h
for parallel version.

In fact, you just have to call MUMPS_distributed_matrix_solve(A,x,b)  on each 
process where A is the local part of your sparse matrix.
MUMPS make the sum of each contribution coming from each process and solve the 
system. Of course, you may need to change some options in the call of MUMPS 
(see the MUMPS documentation).

-- 
Yves.

-------------------------------------------------------------------------
  Yves Renard (address@hidden)       tel : (33) 04.72.43.87.08
  Pole de Mathematiques,                       fax : (33) 04.72.43.85.29
  Institut Camille Jordan - CNRS UMR 5208
  INSA de Lyon, Universite de Lyon
  20, rue Albert Einstein
  69621 Villeurbanne Cedex, FRANCE
  http://math.univ-lyon1.fr/~renard
-------------------------------------------------------------------------